Since its conceptualization in the early 1980s, transformational leadership (TL) has been one of the most studied leadership theories over the past few decades (Mhatre and Riggio, 2014) and has been frequently compared with transactional leadership (Burns, 1978). Whereas, transactional leadership purpose was short-term self-interest and sole exchange of values between the leader and the follower, TL was based on the leader engaging and creating a connection with the followers to increase motivation and morality, and thus, transforming both the leader and follower (Burns, 1978; Northouse, 2018). Future work built upon this idea of the transactional-transformational model of leadership by developing a new continuum known as the full range of leadership model that had three elements ranging from TL to transactional leadership, and ultimately, absence of leadership (or laissez-faire; Bass, 1985).

TL was defined as a leadership form that aims to “achieve follower performance beyond ordinary limits” (xiii) and was proposed to be represented by four dimensions: idealized influence, inspirational motivation, intellectual stimulation, and individualized consideration (Bass, 1985). Idealized influence refers to the leader's ability to serve as a positive model for the follower through high moral standards and ethical conduct, which conveys an ideological vision to the follower along with a sense of collective purpose (Bass, 1998; Bass and Riggio, 2006; Mhatre and Riggio, 2014). Inspirational motivation refers to the leader's ability to inspire and motivate followers by visioning high-performance expectations while simultaneously boosting their confidence to reach them (Bass, 1998; Bass and Riggio, 2006; Mhatre and Riggio, 2014). Intellectual stimulation refers to the leader's ability to stimulate followers' creativity and innovation through challenging tasks as the leader conveys to the follower that they are empowered and trusted. Lastly, individualized consideration refers to the leader's ability to be responsive to the followers' needs and to be able to provide guidance, supports, and mentorship to further enhance their performance and potential (Bass, 1998; Bass and Riggio, 2006; Mhatre and Riggio, 2014).

The theory of authentic leadership (AL) was ignited with the conceptualization of Luthans and Avolio (2003) where AL was viewed as a positive form of leadership instead of negative or inauthentic (Seeman, 1960). Such conceptualization stimulated further research efforts to develop AL from an organizational perspective (Avolio et al., 2004a; Avolio and Gardner, 2005; Gardner et al., 2005). Authentic leaders were defined as those who have a high sense of awareness when it comes to their thinking and behavior and are perceived by themselves and others as being so; moreover, they are fully aware of the context they operate in and have high levels of confidence, optimism, hope, resiliency, and moral character (Avolio et al., 2004b). The definition of AL was further refined to reflect a set of leader behaviors that emphasize both positive psychological capacities and positive ethical climate to foster greater leader self-awareness, internalized moral perspective, balanced processing of information, and relational transparency when working with followers, thereby, fostering positive self-development (Walumbwa et al., 2008, p. 94); the above definition reflects the four dimensions of AL: self-awareness, relational transparency, balanced processing, and internalized moral perspective.

The self-awareness dimension of AL refers to the leader's ability to be aware of their own emotions, values, goals, weaknesses, and strengths (Gardner et al., 2005; Ilies et al., 2013). Moreover, it includes insight or knowledge of “one's multifaceted nature of the self,” which can be gained through exposure to others and being aware of one's impact on them (Walumbwa et al., 2008, p. 95). Relational transparency refers to the leader's transparency in sharing information and expressing their true emotions and thoughts to others without faking or distortion (Walumbwa et al., 2008). An authentic leader is shown to others in an open, honest, yet appropriate manner, which promotes trust in relationships (Kernis, 2003; Wong and Cummings, 2009; Zhang et al., 2021). Balanced processing is a dimension that reflects the leader's ability to keep an objective mindset when making decisions while simultaneously considering others' opinions and thoughts (Walumbwa et al., 2008). Lastly, internalized moral perspective refers to the leader being able to self-regulate through internal moral standards that are not affected by group, organizational, or societal pressures (Walumbwa et al., 2008). In sum, authentic leaders can make decisions consistent with their internal values and moral standards (Walumbwa et al., 2008). The work of Walumbwa et al. (2008) is only one out of many other efforts to conceptualize AL (see Gardner et al., 2011 for a review).

Public corporate scandals over the past decade, such as Enron, WorldCom, Nortel, AIG, and Lehman Brothers, have directed a high amount of attention toward the ethicality of corporate decision-making and leadership. Ethical leadership was defined as personal and interpersonal actions that demonstrate normatively appropriate conduct and promote such conduct to followers through two-way communication, reinforcement, and decision-making (Brown et al., 2005, p. 120). In other words, an ethical leader is a leader who is fair, honest, trustworthy, and principled decision-makers that behaves ethically in his/her personal and professional lives (Brown and Treviño, 2006; Charoensap et al., 2019). Moreover, ethical leaders do not only play the role of moral people by behaving ethically, but they also become moral managers such that they act as role models to influence followers to behave ethically as well (Brown and Treviño, 2006; O'Keefe et al., 2019). Two theories can rationalize how ethical leaders influence the followers' ethical behavior and standards: social learning theory (Bandura, 1977) and social exchange theory (Blau, 1964).

Social learning theory refers to the idea that individuals pay attention to what is attractive in the environment, such as context and other individuals, for guidance (Bandura, 1977). Ethical leaders act as attractive role models to learn from due to their fairness, credibility, and trustworthiness (Brown and Treviño, 2006). Consequently, followers pay attention to the behaviors of the ethical leader to understand what is normatively acceptable and what is not. Moreover, learning gets reinforced when the leader applies rewards and punishments to influence ethical behavior (Brown and Treviño, 2006). Social exchange theory states that social exchange tends to result in feelings of gratitude, trust, and personal obligation, which is different from the transactional exchange as they are contract-like (Blau, 1964, p. 94; Homans, 1961). Therefore, due to the fairness, honesty, and caring of the ethical leader, the followers of the ethical leader are more likely to perceive themselves in a social exchange or personal obligation to reciprocate and “go above and beyond the call of duty for these leaders” (Brown and Treviño, 2006, p. 607). To summarize, ethical leaders tend to influence the followers' ethical behaviors directly through role modeling and indirectly through creating an environment of reciprocity and social exchange (Bedi et al., 2016).

Although servant leadership (SL) was introduced earlier than TL (Greenleaf, 1970), it had not attracted much attention until recently (Liden et al., 2008; Dinh et al., 2014; Lumpkin and Achen, 2018; Eva et al., 2019). Greenleaf's (1970) proposed that servant leaders' highest priority is serving the others' needs and are stimulated by a natural feeling that turns into a conscious choice of leading through serving: “the servant-leader is servant first” (p. 13). Since then, numerous researchers have built upon Greenleaf's (1970, 1977) work (Liden et al., 2008; Spears, 2010; van Dierendonck and Nuijten, 2011). Servant leaders have goals that are not based on self-interest as they aim to form long-term relationships with their followers that extend outside the organization instead of treating organizational goals as their end goal (Ehrhart, 2004; Liden et al., 2008). Eight dimensions were proposed for servant leadership: emotional healing, creating value for the community, conceptual skills, empowering, helping subordinates grow and succeed, putting subordinates first, behaving ethically (Liden et al., 2008).

Although Greenleaf's (1970) definition of SL has been consistently referenced (e.g., van Dierendonck, 2011), it defined SL in terms of its outcomes instead of its behaviors, thus, affecting its utility as a construct definition (Lemoine et al., 2019). Therefore, more recent efforts proposed a refined definition (Eva et al., 2019): an other-oriented approach to leadership that prioritizes the others' individual needs and interests within the organization and the larger community (p. 114). Furthermore, three features of SL were emphasized: motive, mode, and mindset (Eva et al., 2019). The motive aspect indicates that a servant leader has the personal motivation to take responsibility for others instead of focusing on advancing their own agenda or ambitions. The mode aspect emphasizes that the relationship between a servant leader and each follower should be unique, reflecting each follower's different needs, desires, interests, and goals. Lastly, the mindset aspect refers to the servant leader's deliberate focus on follower development, becoming servant leaders (Eva et al., 2019).

The problematic overlap between the NGL forms has been frequently emphasized in the leadership literature (e.g., Anderson and Sun, 2017; Banks et al., 2018; Lemoine et al., 2019). For instance, TL has been suggested to overlap with SL as they both emphasize the need for vision and influence (Bass, 2000, p. 33) by encouraging others to visualize the organization's future and offering compelling reasons to take action (Barbuto and Wheeler, 2006, p. 319). Although some literature argue that the prosocial vision of SL is different from that of the organizational vision of TL (Stone et al., 2004; van Dierendonck, 2011), other efforts propose that TL can be prosocial as well, thus, benefitting the members of the organization alongside meeting the organization's goals (Grant, 2012).

AL and TL appear to overlap a great deal as well. Multiple areas of shared aspects between the two theories, such as a leader's self-awareness and regulation processes, were discussed (Avolio and Gardner, 2005). What made AL unique as a leadership theory was that it promoted positive organizational behavior states of optimism, resiliency, and hope (Avolio and Gardner, 2005; Anderson and Sun, 2017). However, recent research indicates that these psychological resources are related to both TL and AL (Peterson et al., 2009; Gardner et al., 2011). Until recently, the relationship between these two leadership theories has been ambiguous and not adequately distinguished over the years (Neider and Schriesheim, 2011) as AL is sometimes measured using TL items from the MLQ scale (Jensen and Luthans, 2006).

In addition, EL has some key differences from other leadership styles, such as TL and AL (Brown and Treviño, 2006). For instance, the focus of EL is moral management and other awareness; on the other hand, TL emphasizes vision and intellectual stimulation, whereas AL emphasizes self-awareness (Brown and Treviño, 2006). However, key similarities between these leadership styles were also discussed: role modeling, integrity, ethical decision-making, and altruism (Brown and Treviño, 2006). Furthermore, transformational leaders require a strong moral foundation that are central to EL (Bass and Steidlmeier, 1999). AL also shares commonalities with EL regarding principles and ethics (Begley, 2001; Gardner et al., 2011). For example, one of the defining characteristics of authentic leaders is that they “lead from conviction in pursuit of a value-based mission or cause” (Shamir and Eilam, 2005, p. 397). Simply put, authentic leaders tend to have a deep sense of self, values, and beliefs, which they convey to others in terms of principles and ethics (Avolio and Gardner, 2005).

SL has also been linked to AL as authentic leaders try to serve others more effectively by showing a genuine desire to understand their own leadership behaviors, which is similar to the aspect of serving others in SL (Anderson and Sun, 2017, p. 96). Furthermore, a recent review of the empirical body of the moral leadership forms (i.e., servant, ethical, and authentic) by Lemoine et al. (2019) found more commonalities than differences between the constructs, represented by the prediction of expected outcomes, the utilization of common theories, and the investigation of common composition (or measurement overlap). For instance, servant and ethical leaders are suggested to show concern for followers, servant and authentic leaders are suggested to enhance followers' personal growth, and ethical and authentic leaders are suggested to have moral consistency (Lemoine et al., 2019). Moreover, all three moral leadership forms are suggested to demonstrate some form of moral/ethical behaviors (Lemoine et al., 2019).

In sum, although there might be some unique key differences between the NGL forms, the current state of the leadership literature and empirical body suggests a much larger overlap, which can be problematic (Banks et al., 2018; Lemoine et al., 2019). Therefore, the purpose of present analysis is to investigate the question of empirical redundancy and the overlap between the four NGL forms through an integrative meta-analytic examination based upon the theoretical commonalities discussed in this article and other works (e.g., Hoch et al., 2018; Lemoine et al., 2019) in order to offer insight for future directions in the leadership literature.

The studies for the present analyses were collected through a systematic computer-based search of the four leadership styles (authentic, servant, ethical, and transformational). Several databases and search engines were utilized in the process (e.g., ABI-Inform Complete, ProQuest Research Library, ProQuest Digital Dissertations, JSTOR, Business Source Complete, Sage Journals, Google, and Google Scholar). Manual searches were conducted of specific conference proceedings and programs such as the Academy of Management Proceedings. The following key words were used in the search for the studies of interest: servant leadership, ethical leadership, authentic leadership, transformational leadership, authentic, ethical, servant, transformational paired with correlation, and/or correlation matrix, and/or reliability. Reference lists of articles on these four leadership styles were also compared and searched through (i.e., Ng and Feldman, 2015; Banks et al., 2016, 2018; Hoch et al., 2018). The cutoff date for the collected studies was October 2018. Because of the number of studies and observations used in Hoch et al. (2018) analysis of the TL literature, no new meta-analytic aggregation of TL relationships was conducted as the number of studies and observations used in prior studies were sufficiently large to have confidence in the stability of the results (Hoch et al., 2018). That being said, an exception was made for (1) intention to leave (see Table 1), as this relationship has not been examined in prior meta-analytic research, and (2) the relationship between transformational leadership and the other three leadership forms, as the number of studies and observations in prior analyses in the literature were somewhat low and gathering additional studies for these relationships would be critical to the relative weight analyses.

For the studies to be included in the meta-analysis, several established criteria had to be met. First, the studies had to have a working sample (including nurse samples and teacher-principal dyads)—non-working or student samples were excluded. Second, the studies had to report the sample size, the correlation between the leadership style and the variable of interest, and whether measures were self- or other-reports when applicable. Third, non-English studies were translated to English. Fourth, studies that were done only at the group level of analysis were excluded. For the purpose of consistency, only supervisor reports for job performance and OCB were used as supervisor reports tend to be the standard in leadership studies. On the other hand, only self-reports for CWB were used as this is the most common data collection method for this construct and the most consistent with the covert nature of the majority of these types of behavior.

Two authors independently collected a subset of studies, and the overall list was revised by both authors for coding errors and discrepancies. The following information was collected on each study: scales used, their reliabilities, data source (multi/single source), sample size, and correlations of interest. For studies that did not report the reliability coefficient of the scales, an average of the scale reliabilities was computed following the steps of Banks et al. (2016). Furthermore, for studies that reported only the correlations of the related dimensions of the constructs without the overall one, an average of the dimension correlations was computed following the steps of Hoch et al. (2018).

Schmidt and Hunter (2014) random-effects model was used to conduct this meta-analysis due to its predominance in the literature and accuracy (Hunter and Schmidt, 1990, 2004; Hall and Brannick, 2002). This method allows the generation of sample-weighted estimates of the population criterion by correcting sampling errors and attenuation due to unreliability in the criterion and predictor (Fuller and Marler, 2009). In order to accurately interpret the results, a two-step process outlined by Whitener (1990) was utilized. The first step includes investigating whether moderators are present in the data. The second step utilizes the information from step 1 in order to measure the accuracy of the estimated population mean effect size. Based on that, it was determined whether the group of studies involved in this meta-analysis for each relationship measured was from one population (homogenous, indicating the absence of moderation) or from more than one population (heterogeneous, indicating the presence of unidentified moderators). Furthermore. the ubiquitous 75% rule was utilized to provide an indication of heterogeneity/homogeneity. This “omnibus” test is considered a better assessment of homogeneity than significance tests (Hunter and Schmidt, 2004). 95% credibility intervals were also calculated, as suggested by Whitener (1990), to provide another assessment for the presence of moderators (i.e., having zero in the interval or being too wide indicates heterogeneity).

Furthermore, 95% confidence intervals were calculated (Whitener, 1990) to account for the significance of the estimates mean effect size; confidence intervals provide a range of where the population means effect size can be located based on the observed data (Lipsey and Wilson, 2001). Further, in keeping with the best meta-analytic practices and the distortion effect large sample-size outliers may have on meta-analytic parameter estimates (see Fuller and Hester, 1999), a sensitivity analysis was conducted when large sample-size outliers were present in a group of studies. Similar to Balwant (2016), Z scores were calculated (i.e., “Critical Ratio Z” test; Hunter and Schmidt, 2004) to indicate the extent to which the estimated effect size estimates are different from effect size estimates of prior meta-analyses.

Finally, relative weight analysis was conducted using RWA-Web (Tonidandel and LeBreton, http://relativeimportance.davidson.edu) to assess the relative importance of the four NGL variables as predictors of the outcomes included in the aforementioned meta-analyses (Tonidandel and LeBreton, 2011). Relative weight analysis provides accurate estimates of the predictor variables' relative importance when those predictor variables are correlated, as they are in this case (Tonidandel and LeBreton, 2015, p. 208). Since the relative weights generated for each predictor are scaled in terms of variance explained, these values can, therefore, be interpreted as relative effect size estimates (Tonidandel and LeBreton, 2015). Consequently, these values are useful in making determinations of the practical utility of a variable or variables (Tonidandel and LeBreton, 2011).

For the sake of consistency, Hoch et al.'s (2018) results for most of the TL relationship data (i.e., engagement, job satisfaction, organizational commitment, trust in leader, LMX) were used, as it is the only study that analyzed all four NGL forms. However, because (a) research has shown that self-reports tend to exhibit larger effect sizes than other reports (e.g., Fuller et al., 1996; see also current results for EL-OCB) and (b) personal communication Hoch et al. (2018) combined self- and other-reports for their job performance and OCB analyses, Wang et al.'s (2011) results were used for job performance and OCB values as they separated out self- from other-reports. The correlation values reported in Tables 1–4 were used as inputs to this analysis. The following were used as dependent variables in the relative weight analysis: job performance (other-reports), OCB (other-reports), CWB (self-reports), LMX, organizational commitment, job satisfaction, work engagement, intention to leave, and trust in supervisor. For the purposes of comparison with Hoch et al. (2018), a comparison was conducted of the R² computed for the combined NGL forms (the combined model) with the R² computed for TL.